Forecasting Engine and Airframe Development Cost

Development programs for aircraft engines (272 programs) and airframes (338 programs) were evaluated with step-wise regression analysis to find correlations with program cost; i.e., the coefficients of parametric models. Data covering the last 50 to 60 years were included to obtain robust log-log models with approximately 20 independent variables. Standard errors (S.E.E.) less than 0.10 (log megabucks), and multiple correlation coefficients (R-Square) greater than 0.995 were obtained, validating the use of data from programs in the distant past. Some unexpected influence coefficients were found for program cost drivers.